Forthcoming in: M. Lee and M.

Schiefsky (eds), From Refutation to Assent: Strategies of

Argument in Greek and Roman Philosophy (Oxford: OUP) expected 2013.

Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

Susanne Bobzien

In her important 1979 paper “Aristoteles über Syllogismen ‘aufgrund einer

Hypothese’”, 1 Gisela Striker provided an in-depth analysis of these vexing and
perplexing Aristotelian arguments. She showed that they were Aristotle’s way of
providing a logical vehicle for inferences based on other than term-logical relations;
that the ‘hypothesis’ that gives them their name is best understood as a rule that is not
based on a relation of terms, 2 rather than as a premise; and that in those arguments “the
thing taken instead” (τò μεταλαμβανóμενον) is an assertion that is used instead of the
demonstrandum. 3 It was upon reading this paper that I realized that the (then) prevalent
interpretation of later ancient texts on hypothetical syllogisms as presenting the Stoic
theory of indemonstrables needed a thorough revision. For it is possible to show that,
starting from Aristotle’s “syllogisms based on a hypothesis”, a specifically Peripatetic,
(and from the third century CE also partly Platonist) development can be traced through
the centuries up to the late ancient passages on hypothetical syllogistic in Philoponus
and Boethius. Although the Stoic indemonstrables undoubtedly played a role in this
development, the various theories of hypothetical syllogisms over the centuries are all
Peripatetic (and sometimes a little Platonist) in form, function and terminology. 4 The
present paper looks at Alexander of Aphrodisias’ role in this development.

Alexander’s commentaries on Aristotle’s Organon are valuable sources for both Stoic
and early Peripatetic logic, and have often been used as such – in particular for early
Peripatetic hypothetical syllogistic and Stoic propositional logic. 5 By contrast, this
paper explores the role Alexander himself played in the development and transmission
of those theories. There are three areas in particular where he seems to have made a
difference: (1) First, he drew a connection between certain passages from Aristotle’s
Topics and Prior Analytics and the Stoic indemonstrable arguments, and, based on this
connection, appropriated the Stoic indemonstrables as Aristotelian. (2) Second, he
developed and made use of a specifically Peripatetic terminology in which to describe
and discuss those arguments – which facilitated the integration of the indemonstrables
into Peripatetic logic. (3) Third, he made some progress towards a solution to the
problem of what place and interpretation the Stoic third indemonstrables should be
1
Striker [1979]. See now also Striker [2009], 174-8, 201, 237- 238.
2
Striker [1979], 46.
3
Ibid., 43.
4
Here I am in agreement with Maroth [1989]. The arguments the Stoics called hypothetical syllogisms
were completely different from those the Peripatetics called hypothetical syllogisms: see my [1997].
5
E.g. Mueller [1969], Frede [1975], Goulet [1978], Barnes [1984], [1985], Mignucci [1993],
Ierodiakonou, [1990], Bobzien [1996], Speca [2001].
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

given in a Peripatetic (and Platonist) setting. 6 Before I discuss these points in detail,
here are some general remarks about Alexander and the context in which his
contribution to the development of a Peripatetic theory of hypothetical syllogistic
should be seen.

Alexander, like his older contemporary Galen and the Middle-Platonists, was faced with
the Stoic five kinds of indemonstrables and with a rudimentary early Peripatetic theory
of four types of hypothetical syllogisms, both seemingly covering the same logical
ground. 7 The Stoic theory and the terminology that came with it was generally known
and taught, and parts of it at least had become standard logic and standard terminology,
learned, used and/or referred to by members of all philosophical schools in the 1st and
2nd century CE. 8 The early Peripatetic ‘theory’ was known to Galen, and the fact that
some fragments of it are also preserved in Alexander, and in some later texts, suggests
that later Peripatetics were familiar with it. Galen, despite displaying a preference for
the Peripatetic-Platonist approach to logic, remains eclectic in his Institutio Logica and
does not officially side with any school. 9 By contrast, and not surprisingly, Alexander
always defends the Aristotelian or Peripatetic line and habitually attacks the Stoics. Yet
this had not been the manner of all Peripatetics: in the first century BC, Boethus appears
to have adopted the Stoic indemonstrables wholesale, terminology and all (Galen,
Inst.Log.7.2, see below Section 2.2). Propriety in matters of logic – as in other areas of
philosophy – seems to have developed, together with a more historico-philologically
orientated study of the texts of Plato and Aristotle, only after the turn of the millennium.
The competition between the philosophical schools (and various other factors, no doubt)
led to the desideratum that each of the founding philosophers have a view or dogma (if
not a theory) on every philosophical subject matter – including those which had seen the
light of day only after their death. These include prominently fate, the criterion of truth,
and propositional logic; all three philosophical standard topics in the 2nd century CE,
and all three introduced into philosophy only in Hellenistic times, and not originally
part of Plato’s or Aristotle’s philosophy. Thus, Alexander, or some recent predecessors
of his, patched together a ‘theory of fate’ for Aristotle from several of Aristotle’s
writings, and in [Plutarch] On Fate we find a similar patchwork for Plato. 10 In Ptolemy,
we find a Peripatetic ‘theory of the criterion’, 11 in Alcinous (Didasc. ch.4) a Platonist
one, pieced together from excerpts from Plato’s dialogues. Similarly, as this paper
intends to show, Alexander, or some recent predecessors, 12 purposely credited Aristotle
with a theory that corresponds to Stoic propositional logic (i.e. with a hypothetical
syllogistic).

6
For Alexander’s view on the so-called ‘wholly hypothetical syllogisms’ see Bobzien [2000].
7
For the Stoic indemonstrables see e.g. Frede [1974], Bobzien [1996]. For the early Peripatetic theory
see Barnes [1985], Bobzien [2002a], [2002b].
8
Cf. e.g. S.E.P.H.II.157-59; Gal.Inst.Log.6.6; Cic.Top.12.53-14.57, Fin.IV.19.54-5; Plutarch, De E apud
Delphos 386E-387C; Philo De Plantatione 115; Alcinous, Didasc.ch.6; Apul.Int.191.6-11, 201.4-11,
209.9-14, 212.10-12; Aulus Gellius, N.A.XVI.8.1-8; Boethus Perip. acc. to Gal.Inst.Log.7.2; Lucianus
Vitarum Auctio 24; [Galen] Hist.Phil.15.
9
Bobzien [2004].
10
Cf. Alexander On Fate 165.14-171.17 [Plutarch] On Fate 568b-574e.
11
See Mark Schiefsky, this volume.
12
In what follows, I suppress the adjunct ‘or some recent predecessors’; but readers should keep in mind
that it is virtually impossible to decide whether a certain innovation is Alexander’s or whether he adopted
it from lost Peripatetic commentators of the previous generation.

2
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

Unlike in the case of fate, we have no extant separate treatise by Alexander on

hypothetical syllogistic. 13 We have to rely on some remarks and brief discussions in his
commentaries on Aristotle’s Prior Analytics and Topics. Alexander likely also
considered some material relevant to hypothetical syllogistic in his lost commentaries
on the Categories and De Interpretatione. 14 So there is no evidence that Alexander ever
attributes a worked-out theory of hypothetical syllogisms to Aristotle. Rather, what is
remarkable is that Alexander persistently (if not always consistently) interprets and
presents passages from Aristotle’s Organon in a light that makes it appear as if Aristotle
was in the possession of a Peripatetic correlate to the Stoic theory of indemonstrables.

1. The connection of passages from Aristotle’s Organon with the Stoic theory of
indemonstrables and the appropriation of the latter as Aristotelian
Alexander draws two connections between Aristotle’s logic and the Stoic
indemonstrables: first, a connection with Aristotle’s ‘syllogisms from a hypothesis’, and
second, a connection with two of Aristotle’s topoi.

1.1 Aristotle’s syllogisms from a hypothesis and the Stoic indemonstrables

There are four passages germane to the first connection, all in Alexander’s Analytics
commentary. 15 In each of the four passages, Alexander either states or implies that the
indemonstrables are a subclass of Aristotle’s ‘syllogisms from a hypothesis’. In one, in
addition, he actually undertakes to show this. First, his mere claims and suggestions that
there is a connection: (italics mine)

(1) To those ‘by means of another hypothesis’, as he (i.e. Aristotle, An.Pr.41a40f) 16

said, presumably also belong the arguments which are the only ones that the
more recent <philosophers> want to call syllogisms. These are the arguments
that come to be by means of the mode-forming <premise>, as they say, and the
co-assumption, the mode-forming premise being either a conditional or a
disjunction or a conjunction. (Alex.An.Pr.262.28-31) 17

(2) … or, after he (i.e. Aristotle) has said which of the hypotheticals clearly fall
under the presented method (these are both the <arguments> through the
impossible, and the <arguments> in accordance with that-which-is-taken-instead
(metalēpsis) – which include all of the so-called indemonstrables – and the
arguments in accordance with a quality) … (Alex.An.Pr.326.3-5)

13
The view until recently held by the majority of scholars that Ibn Sina (Avicenna) provides evidence for
the existence of a treatise by Alexander on hypothetical syllogisms has been laid to rest by Street [2001].
14
Thus, in Ammonius’ and Al Farabi’s De Interpretatione commentaries we find classifications of
hypothetical propositions or premises, and Al Farabi’s short Categories commentary contains a
classification of consequence (ἀκολουθία) and conflict (μάχη).
15
Alex.An.Pr.262.28-265.5, 326.4-5, 386.27-30, 389.31-390.1, 3-6. Potentially relevant: An.Pr.325.37-
326.1 and 386.22-3.
16
Mueller [2006] 136 n.33 reads εἶπον instead of Wallies’ εἶπεν, at 262.28 and takes Alexander to refer
to the earlier passages Alex.An.Pr.262.9 and 262.28-264.31.
17
For reasons of word limit, and with some regret, I refer the reader to the TLG (Thesaurus Linguae
Graecae) for the Greek text from Alexander.

3
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

(3) (i) Having talked about the <arguments> from agreement and the ones that lead
to the impossible, Aristotle says that there are many other arguments that also
conclude from a hypothesis; he defers speaking about these with more care. … .
(ii) He would mean the hypothetical <arguments> by means of that-which-
connects, which is also called a conditional, and the co-assumption, and the
ones by means of that-which-divides, that is disjunction, and perhaps the ones
by means of the negation of a conjunction. (Alex.An.Pr.389.31-390.1, 390.3-6)

(4) The <arguments> which prove something by leading to the impossible, too, are
“from a hypothesis” … . <This is> so also in the case of the hypothetical
<arguments> by means of that-which-connects, and similarly <in the case of the
hypothetical arguments> by means of that-which-divides. That which has been
posited is not accepted by means of syllogisms but because of the hypothesis;
the syllogism is of something else. 18 (Alex.An.Pr.386.22-23, 27-30)

It appears that the clauses in italics were all used by Alexander to denote the same types
of argument. He refers to them in different ways, but his glosses in passage (3) make it
clear that he intends the same argument types each time. The Stoic terminology in (1),
(2), and (3) (‘mode-forming’, ‘co-assumption’, ‘conditional’, ‘disjunction’, ‘negation of
a conjunction’, ‘indemonstrable’) and the reference to the more recent philosophers
suggest that these are the arguments the Stoics call indemonstrables. (The terminology
is discussed in detail in Section 2.) On this assumption that Alexander intends the same
types of argument each time, taken together the passages suggest he envisages a
classificatory scheme of Aristotle’s syllogisms from a hypothesis as sketched in Figure
1. 19

Figure 1
syllogisms
⁄ | \
deictic syllogisms syllogisms from a hypothesis others?
⁄ | \
from agreement through the impossible the others
⁄ | \ \
with metalēpsis from analogy with quality still others 20
⁄ \ |
‘indemonstrables’ others? 21 those from
⁄ | \ ⁄ | \
conditional disjunctive conjunctive more less the same

If Alexander did not envisage a scheme similar to this, he worked with several
incompatible schemes. (Aristotle’s text is far from clear on the logical relation between
the different types of syllogism he mentions.)

18
For the details of these arguments see Striker [1979].
19
Types with correlates in Aristotle in bold. For syllogisms from a hypothesis see
Arist.An.Pr.A.23,29,44; “through the impossible”: 41a22-37, 45a23-b15, 50a29-38; “from agreement:
50a16-28, 33-5; “with metalēpsis” and “with quality”: 45b15-19; “others”: 41a37-41, 45b15-16, 50a39-
b2.
20
‘Those with some other kinds of hypothetical premises which have been discussed elsewhere’: we
don’t know where Alexander discussed these.
21
From passage (2) it’s unclear whether Alexander thought the so-called indemonstrables exhausted the
syllogisms with metalēpsis.

4
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

Taken literally, Alexander’s classification of the Stoic indemonstrables seems clearly

anachronistic. At the time the Stoics introduced them, Aristotle had been dead for some
time. Should we say that in classifying the Stoic indemonstrables as Aristotelian,
Alexander confuses Peripatetic hypothetical syllogisms with Stoic indemonstrables? 22 I
believe this puts things the wrong way. Rather, Alexander assumed that humans
generally make use of certain ways of arguing or patterns of inference (e.g. we argue
from ‘if Diotima is breathing, then Diotima is alive’ and ‘Now Diotima is breathing’ to
‘Diotima is alive’), and they are justified in doing so, since the resulting arguments are
valid; and that both Aristotle and the Stoics theorized about such arguments: the Stoics
in a way that is at least partially wrong, and Aristotle in the right way. Just that,
unfortunately, Aristotle did not leave in writing a full theory of such arguments. And, as
elsewhere, where Aristotle has not left a worked-out theory or terminology, Alexander
supplies one in – what he takes to be – the Aristotelian spirit. Several factors suggest
that this was Alexander’s approach. Like Galen, he polemicised against what he
regarded as the Stoic ‘formalism’, i.e. their way of determining what kind of argument
or proposition something is by its linguistic form, rather than by its meaning; 23 in his
view, the Stoics systematized the inference patterns at issue in the wrong way and, in
the wake of this, allowed useless arguments to count as syllogisms. However,
Alexander never doubts that there are underlying syllogisms which the Stoics get
wrong, and which Aristotle and the Peripatetics get right. Similarly, he thinks the Stoics
thought – wrongly – that the ‘indemonstrables’ were indemonstrable, whereas Aristotle
did not make this mistake. Again, Alexander believes that the Stoics – wrongly –
thought that the indemonstrables were syllogisms because they are evidently valid basic
(non-demonstrable) arguments, whereas he (in agreement with Aristotle, or so he
assumes) holds that they are syllogisms because they demonstrate that something does
or does not hold (of something). That is why he usually calls them ‘so-called
indemonstrables’ (cf. e.g. passages (2), (6), (7), (10)).

Thus we are made to believe that the arguments the Stoics call ‘indemonstrables’
belong to a class of syllogisms which Aristotle was aware of, but did not himself fully
develop. Alexander fits them into an Aristotelian classification of syllogisms. As part of
this classification, they then automatically sport a number of properties, i.e. all those
which the more generic types of syllogisms of the classificatory scheme share: being
Aristotelian syllogisms, they demonstrate that something holds or does not hold (of
something); 24 they have more than one premise and no redundant premises. (Stretching
his master’s stance to the limit, Alexander also assumes that the so-called
indemonstrables come about through the three figures of the categorical syllogisms and
are brought to completion through the first figure (cf.Arist.An.Pr.A23.)) As hypothetical
syllogisms, the search concerns that-which-is-taken-instead (τὸ μεταλαμβανόμενον
cf.Arist.An.Pr.A29 and below Section 2.2), and “they cannot be led back from the
things hypothesized” (cf.Arist.An.Pr.A44).

22
As suggested by Speca [2001] 52-3, 56.
23
E.g. Alex.An.Pr.373.18-20, 28-35.
24
Unclear whether they are meant to fit Aristotle’s definition of the syllogism.

5
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

By integrating the ‘indemonstrables’ into a presumed Aristotelian taxonomy and

describing them in Peripatetic terms, 25 Alexander makes it look plausible that they are
of Aristotelian origin and thus appropriates them.

In the long passage An.Pr.262-5, Alexander goes beyond merely stating that the
arguments the Stoics call ‘indemonstrables’ are Aristotelian syllogisms from a
hypothesis; he also undertakes to show they are. He proceeds as follows: he provides
examples of several indemonstrables, and, by describing them in the way Aristotle
describes his example of a syllogism from a hypothesis demonstrates that they fit
Aristotle’s description of the “other” syllogisms from a hypothesis (Arist.An.Pr.A44).
In particular, he focuses on the points (i) that the second premise must be proved 26 by a
categorical syllogism; (ii) that the conclusion (the thing-to-be-proved) is established by
means of (διά) the hypothesis, i.e. the first premises; (iii) that this hypothesis in turn
needs no proof, and in fact cannot be proved. 27

Alexander’s argument is unimpressive. First, he presupposes that the second premise of

the indemonstrables is a ‘categorical statement’, with distinct subject and predicate
terms. Thus he excludes existential statements such as ‘providence exists’. Aristotle’s
syllogisms from a hypothesis may have excluded such cases, 28 but the Stoic
indemonstrables certainly did not. Second, Alexander presupposes that the mode-
forming (τροπικά) premises (which he re-interprets as hypotheses) are posited as ‘well-
known’ (γνώριμα). Again, most Stoics would not have agreed, though Aristotle
would. 29 So we have the choice of either charging Alexander with a petitio principii; or
saying that at most he has shown that the indemonstrables, if understood in a Peripatetic
way, tally with Aristotle’s theory of syllogisms from a hypothesis. Either way, the
Stoics would have no reason to think that their indemonstrables have anything to do
with Aristotle’s logic. Presumably this does not matter for Alexander. He gives the
Peripatetics some reasons for believing that the (inference patterns that lie behind the)
Stoic indemonstrables are the sort of thing Aristotle had in mind as some of the “other
syllogisms from a hypothesis”. From a Peripatetic perspective, they do fit, give and
take, the general conception of the “other syllogisms from a hypothesis” that can be
extracted from the Aristotelian passages.

1.2 Aristotle’s Topics and the Stoic indemonstrables

Alexander draws a second connection between Aristotle’s logic and the Stoic
indemonstrables on pages 165-6 and 174-5 of his Topics commentary. Here he connects
the Stoic indemonstrables with some topoi from Aristotle’s Topics. The first passage on
which Alexander comments is from Topics II.4:

(5) One must examine, regarding the point at issue, [i] what is such that if it is, the
point at issue is, or [ii] what is by necessity, if the point at issue is: [i] if one
wants to establish something, one must examine what there is such that if it is,
25
For Alexander’s use of terminology see also Section 2.
26
‘To prove’ is here used not as strictly as in Arist.An.Post.
27
Cf. Alex.An.Pr.264.32-265.13. Still, if it had to be proved, this, too, would have to happen by means of
a categorical syllogism (Alex.An.Pr.263.22-5).
28
See my [2002a].
29
The – probably – Stoic accounts of ‘argument’ in S.E.PH.2.136 and M.7.301-2 suggest that some
Stoics thought the premises in an argument must be agreed upon by the discussants.

6
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

the point at issue will be (for when the former has been proved to hold, the point
at issue will also have been proved to hold); [ii] if, on the other hand, one wants
to refute something, one must examine what it is that is if the point at issue is
(for when we prove that what follows from the point at issue is not, we will have
destroyed the point at issue.) (Arist.Top.111b17-23)

Alexander describes this topos as ‘twofold, since it can both establish and refute from a
consequence'. 30 He considers each sub-type of the topos separately, starting with a
paraphrase of point [i] from Aristotle’s text. Then he continues:

(6) for if the antecedent, then also the consequent, in accordance with the first so-
called indemonstrable, which establishes from a consequence …
(Alex.Top.165.12-13)

and adds an example. Moving to the second sub-type, he again starts with a paraphrase
of point [ii] from Aristotle’s text. Then he continues

(7) for if the consequence <does> not <hold>, nor <does> the antecedent, in
accordance with the second so-called indemonstrable, which refutes from a
consequence … (Alex.Top.166.11-13)

Again he adds an example. It is unclear, how exactly Alexander envisages the relation
between the ‘twofold’ Aristotelian topos and the first two indemonstrables. He certainly
believed that the two so-called indemonstrables are somehow related to the two ways of
the topos: The topos is said to be ‘establishing and removing from consequence’, and
the first so-called indemonstrable is said to be ‘establishing from consequence’; the
second to be ‘removing from consequence’. Moreover, he seems to think that the so-
called indemonstrables explain the corresponding sub-topos (‘for if …’). However, the
exact purpose of the explanation is hard to gauge. Alexander talks about the so-called 31
first and second indemonstrables. He seems to regard the so-called indemonstrables as
patterns of inference or argument forms or schemata. His formulations (‘for if …, then
also …’; ‘for if …, neither …’) suggest that, these patterns can be used as justifications
of the two sub-topoi. Thus Alexander considers these patterns called ‘so-called
indemonstrables’ as logically prior to the topos. This, again, suggests that Alexander
thinks of certain inference-patterns similar to modus ponens and tollens as existing valid
patterns; patterns Aristotle was aware of and which the Stoics wrongly thought of as
being indemonstrable. This is confirmed by another passage from Alexander’s Prior
Analytics commentary (Alex.An.Pr.336.13-20), in which Alexander couches two
arguments in a Peripateticized version of the Stoic mode-arguments 32, and then adds:

(8) The establishing is in accordance with the second indemonstrable and the
refuting is in accordance with the first. 33

30
Ὁ μὲν τόπος ἐστὶν ἐξ ἀκολουθίας. διπλοῦς δ’ ἐστί· καὶ γὰρ κατασκευαστικός ἐστι καὶ ἀνασκευαστικός.
Alex.Top.165.6-7.
31
The Stoics called indemonstrables the individual actual arguments that satisfied the descriptions of
first, second, etc. indemonstrable. See my [1996] Section 1.
32
λογότροποι, see DL.7.77.
33
διὸ ἡ μὲν ἀνασκευὴ κατὰ τὸν δεύτερον ἀναπόδεικτον ἒσται, ἡ δὲ κατασκευὴ κατὰ τὸν πρῶτον
(Alex.An.Pr.336.18-20).

7
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

Here again the Aristotelian methods of establishing and refuting are said to be ‘in
accordance with’ (κατά) the indemonstrables. The latter are hence considered as
logically prior.

The second passage on which Alexander comments is from Aristotle’s Topics II.6:

(9) In the case of things of which hold one and only one of two <predicates>, as for
instance of a human being holds either illness or health, if we are well-equipped
to argue about the one, that it holds or does not hold, we will also be well-
equipped with regard to the remaining one; this converts with regard to both; for
[iii] when we have proved that the one <predicate> holds, we will also have
proved that the other does not hold; and [iv] when we prove that the one does
not hold, we will have proved that the remaining one holds. (Arist.Top.112a24-
31)

In this case, the connection Alexander draws between the Aristotle passage and the
indemonstrables is more complex. He starts with describing the topos as ‘establishing or
refuting from a conflict’ (τόπον ἐκ μάχης κατασκευαστικόν τε καὶ ἀνασκευαστικόν,
Alex.Top.174.7). The parallel to his treatment of the previous topos is obvious. Taking
his comments on the two passages (Alex.AnPr.165-6 and 174-5) together, we can see
what Alexander did: He combined the Stoic pair of logical expressions ‘consequence’
and ‘conflict’ with the Peripatetic pair of ‘establishing’ and ‘refuting’. It seems that this
combination of the pairs of terms is not known before Alexander. We find it in some
later texts, with slight variations that reflect further developments. 34 So here Alexander
may have been innovative. At any rate, in this passage we have evidence of one
important step towards the incorporation of Stoic logic into Peripatetic and Platonist
syllogistic.

After this description of the topos Alexander interprets Aristotle’s passage in the light
of an Aristotelian logical distinction that originates with the Categories. The distinction
is that of contraries without intermediates, contraries with intermediates, and
contradictory opposites (which are not contraries and have no intermediates). In
Categories 10, Aristotle classifies these as three different kinds of opposites
(ἀντικείμενα), and explains each one. 35 Here Alexander draws another important
connection: the Aristotelian kinds of opposites are understood as three different types of
conflict. Thus a Peripatetic logical distinction and a Stoic logical term are linked. 36

Alexander suggests that in the cases of contraries without intermediates and of

contradictories, the topos under discussion can be used both for establishing and for
34
Galen used this pair of terms for a Peripatetic logical distinction; but he didn’t make the connection
with the Topics; or if he did (in his lost commentaries?) this didn’t survive.
35
Arist. Cat.11b17-23, 11b38-12.a25, 13a37-b35 (cf. Amm.Cat.95.8ff). Aristotle’s talks of contradictory
opposites as ‘things opposed as affirmation and negation’; Alexander uses the expression from the De
Interpretatione, αντίφασις. Aristotle describes the contraries (εναντία) without middle as τούτων οὺδέν
ἐστιν ὰνα μέσον, those with middle as τούτων ἔστιν τι ὰνα μέσον. Alexander and later commentators tend
to use the terms ἔμμεσα and ἄμεσα. Aristotle does not say that contradictory opposites have no middle,
but such a description is in keeping with Cat.13a37-b35.
36
He illustrates the cases of contraries with examples from Arist.Cat.10 and adds further examples from
other Aristotelian works.

8
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

refuting. The connection between Aristotle’s topos and the contraries without
intermediates is obvious, and Alexander appears merely to spell out what Aristotle
would have taken for granted. (Like Aristotle, Alexander only talks about contraries
without combination, i.e. contrary pairs of terms, not contraries with combination, i.e.
contrary pairs of sentences.)

The case of contradictories is also – if somewhat forced – dealt with within term logic:
it holds of everything, if you have proved that something is affirmed of it, then you have
refuted that that thing is negated of it, and vice versa. The topos under discussion can
thus be used both for establishing and for refuting in the case of contradictories, too
(Alex.Top.175.2-10). I am unsure whether this interpretation of the topos would have
found Aristotle’s approval.

Alexander’s introduction of the opposites with intermediates into the topos would most
probably not have found Aristotle’s approval. Aristotle starts from a certain relation of
terms (contraries without intermediates) and explains what inferential moves of
establishing and refuting one can make given this relation. Alexander, by contrast,
seems to have focussed on two general inference patterns that involve conflict. The text
suggests that these two patterns are along the lines of “If A and B conflict, then if we
have A, we have refuted B” and “If A and B conflict, then if we don’t have A, we have
established B”. Starting from these patterns, Alexander investigates the above-
mentioned three Aristotelian types of opposites, as to whether they “fit” them. He
concludes that contradictories and contraries without intermediates fit both patterns, but
contraries with intermediates fit only the pattern for refuting. Once more, Alexander
appears to regard the patterns of inference as independent of and prior to the specific
topos under discussion, and as having wider application than the topos. The passage in
Aristotle’s Topics betrays no such assumption of underlying independent inference
patterns. Again, Alexander’s specific interpretation of an Aristotelian passage facilitates
his drawing a connection with the Stoic indemonstrables: the fourth and fifth
indemonstrable show markedly clearer similarities to Alexander’s two inference
patterns than to Aristotle’s Topics passage.

Alexander introduces the indemonstrables at the very end of his comments on the
Topics passage:

(10) And the proof that is fitting for the <contraries> without intermediates is
rather the one by means of the so-called fifth indemonstrable, which is the one
that concludes from that-which-divides and the contradictory to one of the
<components> in that-which-divides the remaining <component>. But for the
contraries with intermediates the <proof> by means of the fourth <so-called
indemonstrable is fitting>, which is the one that refutes from that-which-divides
and one of the <components> in that-which-divides the other <component>.
(Alex.Top.175.21-6)

In this passage, Alexander does not mention the Aristotelian contradictories at all. His
correlation of the two types of contraries with the fourth and fifth types of
indemonstrables seems unique. We discuss his correlation of the fourth indemonstrables
with the contraries with intermediates in Section 3.2. Here we note that what he says
about contraries without intermediates is unexpected, given his earlier remarks on the

9
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

contraries. We would have expected him to say that both the fourth and the fifth
indemonstrable fit the case of contraries without intermediates. Yet he correlates only
the fifth with these. 37

We disregard this problem and focus on Alexander’s positive input. He reports the Stoic
description of the fifth indemonstrables almost correctly, with the exception of some
terminological modification (for which see below and Section 2.2). But the use he
makes of this description is distinctly un-Stoic. The Stoic theory of indemonstrables is a
theory of basic formally valid arguments to which all more complex formally valid
arguments can be reduced, but which themselves need no proof. 38 The Stoics would
simply say that everything with the form described is a fifth indemonstrable. This
includes arguments from false premises and arguments that in the disjunction have the
same component assertible twice. Alexander despises this focus on the mere validity of
arguments, since it produces useless argument forms. 39

What interests Alexander is how one can produce proofs by using valid inference
patterns. Proofs need to have true premises. Alexander (in line with Aristotle’s Topics)
considers first what types of complex premises are such that their instances are always
true. In the case at issue, the relevant types must all fit the general pattern of dividing
premises/propositions. They turn out to be those types of dividing hypothetical premises
in which the two components express opposites: the dividing premises/propositions are
always true if they express contraries with intermediates; or contraries without
intermediates; or contradictories. Second Alexander considers by addition of what true
categorical premises/propositions one can then establish or refute something.

The Stoic fourth and fifth indemonstrables are thus re-interpreted as general inference
patterns and put to a distinctly Peripatetic use within Peripatetic logic: they provide the
general pattern for proofs that refute or establish something from a conflict (μάχη) of
things. Three of Aristotle’s types of opposites are interpreted as three different types of
conflict; each can be used for producing proofs. Thus, again, the indemonstrables are
treated as logically prior to the topos at issue.

The details of this use of the Stoic indemonstrables are messy. Alexander’s commentary
on the topos (Alex.Top.174-6) before the introduction of the indemonstrables is done in
terms of term-logic. Where Alexander introduces the descriptions of the
indemonstrables, he leaves out all nouns that would allow us to pin them down as Stoic
or as Peripatetic:

… the so-called fifth indemonstrable, which is the one that concludes from that-
which-divides and the contradictory to one of the <components> in that-which-
divides the remaining one.

The Stoic version would be:

37
We note that to his credit, Alexander makes his statement sound tentative, using the expression ‘rather’
(μᾶλλον). Of course, the idea that both patterns fit, but one may fit somewhat better than the other is also
not terribly attractive.
38
Cf. my [1996] Section 1.
39
Cf. e.g. Alex.An.Pr. 18.12-22.

10
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

… the fifth indemonstrable, which is the one that concludes from a disjunctive
assertible and the contradictory to one of its disjuncts the remaining disjunct. 40

The Stoic relation of the disjunctive assertible and the assertibles that are used as
disjuncts is such that the truth-value of the disjunctive assertible depends on the truth-
values (and modal relations) of the assertibles used as disjuncts. At the other extreme,
we would have a Peripatetic version that does not acknowledge the existence of
molecular propositions:

… the so-called fifth indemonstrable, which is the one that concludes from a
dividing hypothetical premise and the contradictory to one of the terms in the
dividing hypothetical premise the remaining term as holding of the object in
question.

Various intermediate versions can be constructed. Since in our passage Alexander is

commenting on the Topics, it is more likely here (than in his commentary on the Prior
Analytics) that he considers ‘naked’ relations between terms rather then the relation
between sentences or premises that contain two terms and share one term. But this is
merely conjectural.

What matters are the following points: With the chosen formulations, Alexander does
not commit himself to the elements of propositional logic in Stoic logic. Nothing
suggests that that-which-divides is such that its truth-conditions depend on the truth
(and modal relations) of some truth-bearers that function as its components. There is in
fact no evidence that that-which-divides is thought to have logically independent
components that are truth-bearers. Historically, the passage Alex.Top.174-6 is
important, since it is the earliest that connects the fourth and fifth Stoic indemonstrables
with Aristotle’s Topics (and the only one in Alexander). Moreover, it is the earliest that
connects the Stoic indemonstrables with Aristotle’s theory of opposites. Later
commentators take up both points and improve on them.

To sum up this section: Alexander has given the first two and the last two types of Stoic
indemonstrables a home in Aristotelian dialectic. He has appropriated them as
Peripatetic tools, in the form of argument patterns, for establishing or refuting theses.
He has identified particular Aristotelian topoi which correspond to them, or which are
‘cases’ or ‘examples’ of them. Moreover, he has connected two of the indemonstrables
with Aristotle’s theory of opposites from the Categories. Thus, in a second and third
way, the indemonstrables (four of them, more precisely) have been anchored in
Aristotle’s logic.

2. The development and use of a specifically Peripatetic terminology for the

Stoic ‘indemonstrables’ 41
The greater awareness of the differences in the views of the various philosophical
schools together with the philologically and historically more thorough study of the
original texts of Aristotle and Plato in the second century CE (in contrast with the 1st

40
E.g. DL.7.81, SE.PH.2.158, Gal.Inst.Log.6.6.
41
This section requires knowledge of ancient Greek. It can be skipped.

11
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

BCE) is reflected in Alexander’s choice and use of terminology. Thus, not only does he,
in his comments on hypothetical syllogistic, bring together material from Aristotle,
early Peripatetics, Stoics, and later Peripatetics (including himself); he also preserves
and juxtaposes elements of at least four different (though partly overlapping) sets of
terms:

(a) Although Aristotle had no terminology for hypothetical syllogisms (as he had no
theory of such syllogisms), his general theory of logic, the terminology he used for
syllogisms from a hypothesis, and the terminology used in the topoi discussed in
Section 1.2 taken together provided Alexander with a basic stock of useful
expressions.
(b) The early Peripatetics had made some rudimentary advancement towards a theory
of hypothetical syllogisms and appear to have introduced some basic vocabulary
that went beyond Aristotle’s. 42
(c) The Stoics had a complete set of technical terms for their syllogistic. 43 When Stoic
syllogistic became the logic of hypothetical syllogistic in the 1st century BCE,
many Stoic terms became part of the common terminology used in logic, although
we witness some modifications. 44 Some Peripatetics of the 1st BCE seem to have
adopted together with parts of Stoic theory also parts of Stoic terminology. 45
(d) In the second century CE, there is demand for a distinctly Peripatetic (and also for
a distinctly Platonist) theory of hypothetical syllogistic. One way of marking out a
distinct theory is by introducing a distinct terminology. This seems to be precisely
what Alexander did. His surviving commentaries provide evidence that he
deliberately replaced terms from the Stoic theory of indemonstrables by terms that
are either taken from the closest Aristotelian or early Peripatetic correlates, or are
coined with an eye to expressions from Aristotelian and early Peripatetic logic. For
this purpose Alexander uses primarily two devices. First, he glosses one term by
another, using phrases of the form ‘x, that is, y’ (in Greek ‘x καὶ y’, where καί is
used epexegetically). He generally puts first the less familiar term (that is the
Aristotelian, early Peripatetic, or newly coined one), and explains it with the one
more familiar at his time, usually either the Stoic one, or the common term that
had been developed from the Stoic one. Second, Alexander explicitly distances
himself from the Stoic, or Stoic-derived common, terminology; he does this in
particular where he thinks that the descriptive component of the term is inaccurate.

Alexander’s use of the sets of terms (a) to (d) is not systematic. Nor is it random.
Naturally, he employs Aristotle’s own terms when he presents or paraphrases
Aristotle’s views. Where necessary, he explains them in terms of common or later
Peripatetic terminology. 46 Equally naturally, he uses Stoic terminology mainly when
presenting and criticising Stoic theory. 47 He sometimes makes use of Stoic terms that
have become common coinage where there is no Peripatetic substitute or where he
explains the early Peripatetic equivalents or newly introduced Peripatetic terms.

42
Cf. my [2002a], [2002b], [2004] for details.
43
Cf. my [1996].
44
Cf. Section 2.1 for some examples.
45
Galen is a witness to this development, see my [2004].
46
So for instance at Alex.An.Pr.11.18, 262.32 κατηγορικός.
47
So for instance ἀξίωμα, Alex.An.Pr.177.31, 179.32, 180.2; τò λῆγον at Alex.An.Pr.177.21, 178.28.

12
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

However, he also sometimes uses more recent Peripatetic terms without glossing them
in this way. This last fact suggests that these more recent Peripatetic terms were at the
time already established to some degree among the Peripatetics, or at least among his
students.

If Alexander’s strategy as described in (d), was successful, we should have the Stoic
theory of indemonstrables not only Peripatetically modified and firmly grounded in
Aristotle’s logical system, but also cleverly terminologically disguised in a way that
would suggest to someone historically not on the ball that there has always been a
worked-out Aristotelian theory that covered the same logical ground as the Stoic
indemonstrables. I now consider Alexander’s choice of terminology for hypothetical
syllogistic in detail.

2.1 Propositions or sentences and their components

Alexander uses the Peripatetic term πρότασις, rather than the Stoic term ἀξίωμα
(‘assertible’), for propositions (or meaningful sentences). Like Aristotle and the early
Peripatetics, he also uses πρότασις for ‘premise’. To denote simple propositions,
Alexander uses the adjective ‘categorical’ (κατηγορικός), not the Stoic ‘simple’
(ἁπλοῦς). 48 Thus Alexander uses an Aristotelian term, but in a different meaning than
Aristotle himself; and he is aware of this. 49 (This option must have seemed preferable to
using a Stoic term.) Instead of the Stoic expression ‘mode-forming propositions’
(τροπικός, i.e. those which can be used as a complex premise in a Stoic syllogism)
Alexander uses the Peripatetic ‘hypothetical proposition’ (ὑποθετικὴ πρότασις, which
for the early Peripatetics seems to have meant ‘hypothetical premise’ and not
‘hypothetical proposition’). 50 There seems to be only one passage where Alexander
indubitably uses ὑποθετικὴ πρότασις for hypothetical propositions (Alex.Top.191.18);
in other passages it is unclear whether the intended meaning is proposition or premise
(e.g. Alex.An.Pr.11.17-20, 17.7-8, 324.7, 327.2-3), and when Alexander reports
Aristotle, it is likely intended as premise.

For the various kinds of mode-forming propositions, Alexander appears to introduce his
own terms. In later antiquity, the Stoic terms for conditional and disjunctive assertibles
(τὸ συνημμένον ἀξίωμα and τὸ διεζευγμένον ἀξίωμα) were shortened to τὸ συνημμένον
and τὸ διεζευγμένον and took on a life outside Stoic logic, as independent noun phrases,
not as elliptic for conditional assertible and disjunctive assertible. They were used by
Peripatetic, Platonist, and other non-Stoic philosophers who would call propositions
προτάσεις, not ἀξιώματα. Aristotle had no terms for such non-simple propositions. The
early Peripatetics – if we trust Galen 51 – had words for the two types of relations
between things that are signified or expressed by hypothetical προτάσεις: συνέχεια and
διαίρεσις; and they referred to the corresponding types of hypothetical premises as
προτάσεις κατὰ συνέχειαν and κατὰ διαίρεσιν. In Alexander we occasionally find τὸ
συνημμένον in non-Stoic contexts (Alex.Met.318.23). Much more frequently, we find
that-which-connects (τὸ συνεχές – not ἡ συνεχής πρότασις!). Alexander glosses this
expression several times by τὸ συνημμένον (e.g. Alex.An.Pr.262.7, 390.4-5 and text

48
We find this also e.g. in Gal.Inst.Log.2.2, 6.2. Alcin.Didasc.11.15.
49
See e.g. Alex.An.Pr. 11.18, καλουμένης κατηγορικῆς.
50
See my [2002a] and [2002b].
51
See my [2002b].

13
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

(3)(ii)) Thus we can safely assume that Alexander took the latter term, τὸ συνημμένον,
to be generally understood, and the former, τὸ συνεχές, as in need of explanation. In
Alexander’s commentaries, τὸ διεζευγμένον occurs on its own, without explaining
another term, only in Stoic context. In non-Stoic contexts Alexander uses it for glossing
that-which-divides (τὸ διαιρετικóν; e.g. Alex.An.Pr.20-1). The two terms τὸ συνεχές and
τὸ διαιρετικόν don’t occur before Alexander in logical context. Thus it seems likely that
Alexander himself coined the two expressions in parallel to the original Stoic ones τὸ
συνημμένον and τὸ διεζευγμένον, and intended them as a replacement for the latter.
They could be understood as being truly Peripatetic, since they are formed in analogy
with the early Peripatetic ἡ κατὰ συνέχειαν ὑποθετικὴ πρότασις and ἡ κατὰ διαίρεσιν
ὑποθετικὴ πρότασις. For the Stoic third type of mode-forming assertible and how
Alexander dealt with it, see Section 3 below. Here we only mention the following. The
Stoics, in line with their propositional-logical approach, called these kinds of
propositions conjunction-negating assertible (ἀποφατικὸν συμπεπλεγμένου ἀξίωμα) or
negation of a conjunction (ἀποφατικὸν συμπλοκῆς/συμπεπλεγμένου, S.E.P.H.2.158
M.8.226, D.L.7.80): negation has the largest scope and thus the category of assertibles
to which these belong is (non-simple) negations. We have a compound proposition,
compounded from a negation and a conjunction by prefixing the negation particle to the
conjunction. By contrast, Alexander (like Galen and most later ancient authors) called
their equivalents a negative conjunction (ἀποφατικῆ συμπλοκῆ, below passage (11)).
This, once more, suggests that the propositional-logical element from Stoic logic is lost.
The negative conjunction is just another type of hypothetical premise/proposition from
which inferences can be drawn.

The Stoic terms for the components in a conditional, i.e. the antecedent and the
consequent, were τὸ ἡγούμενον and τὸ λῆγον. Alexander instead uses the Aristotelian
pair τὸ ἡγούμενον and τὸ ἑπóμενον. However, Aristotle himself used this latter pair of
expressions mostly for terms rather than propositions. 52 Thus we have another case
where Alexander chooses an Aristotelian term but doesn’t keep to Aristotle’s main use
of it.

2.2 Syllogisms and their components

The picture is similar for syllogistic. We saw that Alexander associated the Stoic
indemonstrables with Aristotle’s syllogisms from a hypothesis. This fact is reflected in
his terminological choices. Where Alexander uses the name ‘indemonstrable’, he
usually does so only with the addition ‘so-called’ (λεγόμενον). His reason for this is not
just that the name is Stoic, but more importantly, that he thinks it to be inaccurate. The
Stoics, and at least one Peripatetic, Boethus, believed the indemonstrables to be indeed
indemonstrable, i.e. not in need of proof since self-evident. 53 The Peripatetics regarded
some categorical syllogisms as indemonstrable, usually those of the first figure.
Alexander – purporting to follow Aristotle – accepts nothing but the first figure

52
E.g. Arist.An.Pr.43b17.
53
Gal.Inst.Log.7.2. Maroth maintains that Galen reports a discussion within the Peripatetic school, in
which Boethus defended the priority of the hypothetical before the categorical syllogisms (Maroth [1989]
246-7). However, in Galen the terminology and preceding context are Chrysippus' logic, and (pace
Maroth) we have no evidence of the use of the pair of terms "categorical syllogism" and "hypothetical
syllogism" before the 2nd century CE.

14
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

categorical syllogisms as indemonstrable. 54 Hence for the types of arguments the Stoics
call indemonstrables a different name is not just desired but philosophically required.

As the generic term Alexander uses ‘hypothetical syllogism’. This may have been used
already by earlier Peripatetics, or was in any case easily coined by taking Aristotle’s
phrase ‘syllogisms from a hypothesis’ as model. Alexander does not doubt that the
arguments the Stoics call ‘indemonstrables’ (and the later so-called ‘mixed’
hypothetical syllogisms in general) are syllogisms. His justification of this fact is
Aristotelian, however, not Stoic. 55

Alexander calls the first premises of the hypothetical syllogisms, generically,

‘hypothetical premises’. 56 It is less clear whether he had a way of referring to the
specific kinds of premises. For the second premise, which the Stoics called co-
assumption (πρόσληψις), Alexander considers τὸ μεταλαμβανόμενον and τὸ
προσλαμβανόμενον. He takes these to be synonyms, the first being the term Aristotle
and some early Peripatetics used, the second having been used by some early
Peripatetics (perhaps in addition to the first) (An.Pr.19.4-5, 262.6-9. However, whereas
in Aristotle μεταλαμβανόμενον is used to indicate that the premise is that-which-is-
taken-instead, 57 (instead of the conclusion, as that which is to be proved by a categorical
syllogism, that is), 58 Alexander interprets it as meaning ‘changed assumption’, in the
sense that in the hypothetical proposition/premise, the component proposition p, say, is
taken as being hypothesized only, whereas in the ‘changed assumption’, p is taken to
state that something is the case; it has thus changed – changed in its status of force, as
we might say. 59 This may show awareness at Alexander’s time of what is called the
‘Frege point' in contemporary philosophy.

Moreover, unlike the Stoics, Alexander at least sometimes takes πρόσληψις (and also
μετάληψις) to be the act of taking a proposition as the additional premise, and τὸ
προσλαμβανόμενον and τὸ μεταλαμβανόμενον as the result of this act. For the Stoics
the πρόσληψις is an assertible (an incorporeal entity), 60 taken as premise, never the act
of taking the assertible as premise. (Similarly, Aristotle and Alexander sometimes use
‘syllogism’ for the act of deducing (‘deduction’), whereas for the Stoics it always
denotes the argument itself, i.e. a composite of sayables (λεκτά).)

For the five individual types of ‘indemonstrable’ syllogisms, there appears to have been
no set of Peripatetic expressions on offer which Alexander could have adopted
wholesale. To denote them, Alexander uses two different sets of description, each
corresponding to one way he connected them with Aristotle’s Organon (see Section 1).
The first, found in Alexander’s Analytics commentary, is based on his terms for
conditionals and disjunctions (cf. Alex.An.Pr.390.3-5, 386.27-8, passages (3) and (4)
above):

54
E.g. Alex.An.Pr. 24.2-12.
55
Cf. e.g. Alex.AnPr 265.1-24.
56
Cf. Alex. An.Pr.17.8-9, 326.6-7, Top.63.25, 191.18.
57
Μεταλαμβάνω in its meaning of ‘to take instead of’, cf. Striker [1979] 43.
58
See my [2002a].
59
Alex.An.Pr.263.26-33; μεταλαμβάνω used in its meaning of ‘to change’.
60
E.g. D.L.7.76.

15
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

• the hypothetical syllogisms/arguments by means of that-which-connects, which is

also called a conditional, and the co-assumption (οἱ διὰ τοῦ συνεχοῦς, ὃ καὶ
συνημμένον λέγεται, καὶ τῆς προσλήψεως ὑποθετικοὶ <συλλογισμοί/λόγοι>)
• the hypothetical syllogisms/arguments by means of that-which-divides, which is
also called disjunction, and the co-assumption (οἱ διὰ τοῦ διαιρετικοῦ, ὃ καὶ
διεζευγμένον λέγεται, <καὶ τῆς προσλήψεως ὑποθετικοὶ συλλογισμοί/λόγοι>)

These are descriptions in the style of the early Peripatetics, 61 each describing the two
premises through which the syllogism comes about. In a second passage we find
abbreviations of these descriptions: “the hypotheticals by means of that-which-
connects” and “the hypotheticals by means of that-which-divides” (οἱ διὰ τοῦ συνεχοῦς
ὑποθετικοὶ <συλλογισμοί/λόγοι> and οἱ διὰ τοῦ διαιρετικοῦ ὑποθετικοὶ
<συλλογισμοί/λόγοι>). Whether these may actually have functioned as names, rather
than mere descriptions, of these types of syllogisms is unclear. I have not found them in
any author other than Alexander. In any event, these descriptions (or names) are clearly
in the Peripatetic tradition, based on Aristotelian and early Peripatetic terms and
descriptions of what was to become to be the hypothetical syllogisms. 62

Alexander’s second set of descriptions, the one in his Topics commentary, is based on
his terminology for the relevant topoi. Alexander describes the relevant first topos as
τόπος ἐξ ἀκολουθίας κατασκευαστικὸς τε καί ἀνασκευαστικός (Alex.Top.165.6-7,
174.5-6); the first resulting syllogism as the first so-called indemonstrable, which is ἐξ
ἀκολουθίας κατασκευαστικός; and the second resulting syllogism as the second so-
called indemonstrable, which is ἐξ ἀκολουθίας ἀνασκευαστικός. He describes the
second relevant topos as τόπος ἐκ μάχης κατασκευαστικὸς τε καί ἀνασκευαστικός
(Alex.Top.174.6-7); the first corresponding syllogism as the fifth so-called
indemonstrable, which is ἐκ διαιρετικοῦ καὶ τοῦ ἀντικειμένου ἑνὶ τῶν ἐν τῷ διαιρετικῷ
τὸ λοιπὸν συνάγων; and the second corresponding syllogism as the fourth so-called
indemonstrable, which is ἐκ διαιρετικοῦ καἰ τοῦ ἑτέρου τῶν ἐν τῷ διαρετικῷ ἀναιρῶν τὸ
ετερον. 63 Thus it is possible that the fifth was considered as ἐκ μάχης κατασκευαστικὸς,
and the fourth as ἐκ μάχης ἀνασκευαστικός, although Alexander does not explicitly say
so. In any case, Alexander’s description, even though it follows the Stoic standard
definition, has taken on a Peripatetic colouring: we have διαιρετικόν instead of
διεζευγμένον. Moreover, instead of the Stoic τὸ ἀντικείμενον τοῦ λοιποῦ ἔχων
συμπέρασμα we have the Peripatetic ἀναιρῶν τὸ ἕτερον; that is, instead of being
formulated in terms of contradictoriness, the conclusion is formulated in terms of
refuting. Alexander is our earliest source for these features, and they all recur, further
refined, in later Aristotle commentators. The fourfold descriptions of the
‘indemonstrables’ in terms of the two pairs of expressions ‘consequence’ and ‘conflict’
and ‘establishing’ and ‘refuting’ seem to be echoed in one text only, namely Alcinous’
Handbook of Platonism (Didasc. ch.6). Alcinous’ dates are uncertain, but he is likely to
have been either a contemporary of Alexander or a generation or so later.

61
Cf. Alex.An.Pr.262.31-2 μίκτους ἐξ ὑποθετικῆς προτάσεως καὶ δεικτικῆς together with my [2002a].
62
See also Alex.An.Pr.19.4 ὁ <ἐκ αντιφάσεως> διαιρετικὸς συλλογισμός and Alex.Top.11-12, 175.
63
Elsewhere he classifies a different kind of argument as fifth indemonstrable, see Section 3.

16
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

I conclude the present section with a table of the various expressions relevant to
hypothetical syllogistic found in Alexander (or which he is likely to have known),
sorted according to ascription and/or likely origin (Figure 2).

Figure 2
Aristotle early Peripatetics 64 Stoics Alexander
proposition / statement proposition simple assertible categorical proposition
? ? mode-forming proposition hypothetical proposition
- ? conditional assertible that-which-connects
- ? disjunctive assertible that-which-divides
- ? conjunctive assertible conjunction?
- ? negation of conj. assertible negative conjunction?
antecedent antecedent antecedent antecedent
consequent consequent ‘ending’ consequent
‘following’ connection to follow consequence
?’conflict’ division to conflict conflict
(syllogism from hypoth.) syll. from hypothesis indemonstrable syllogism hypothetical syllogism /
so-called indemonstrable syllogism
- in acc. w. connection first indemonstrable hyp.syll. through that-which-connects /
establishing from consequence
- in acc. w. connection second indemonstrable hyp.syll. through that-which-connects /
removing from consequence
- ? third indemonstrable (See Section 3)
- in acc. w. division fourth indemonstrable hyp.syll. through that-which-divides /
establishing from conflict
- in acc. w. division fifth indemonstrable hyp.syll. through that-which-divides /
removing from conflict
that-which-is-taken-instead that-which-is-taken-instead co-assumption changed premise
- that taken in addition co-assumption that taken in addition
- proving premise (co-assumption) categorical premise
(hypothesis) hypothetical premise mode-forming assumption hypothetical premise

Figure 2 shows that Alexander has succeeded in building up a vocabulary for

hypothetical syllogistic that is in its entirety non-Stoic and based on Aristotelian and
early Peripatetic terms.

3. The problem of the negative conjunctions and the third indemonstrables

Historically perhaps the most fascinating aspect of Alexander’s appropriation of the
Stoic indemonstrables for Peripatetic purposes is his treatment of the third
indemonstrables and their mode-forming premises, the conjunction-negating assertibles.
We know from Galen that there was a debate about the usefulness and validity of the
Stoic third indemonstrables, and also that they had no early Peripatetic correlate. 65
Galen himself rejected negative conjunctions with a contingent relation between the
conjuncts (e.g. ‘Theo walks and Dio talks’) as unsuitable for producing syllogisms. But
he considered negative conjunctions with partially incompatible conjuncts (i.e. those
which cannot be true together, e.g. ‘Theo is in Athens and Theo is on the Isthmus’)
suitable for syllogistic. He took them to express a relation of incomplete conflict and
called them quasi-disjunctions (παραπλησία διεζευγμένοις), despite their linguistic form
of a negative conjunction. 66 Later Aristotle commentators reserved the ‘third mode of

64
Maroth's conjecture that the early Peripatetics used the word hypothetical (ὑποθετικός) for propositions
and syllogisms ([Maroth 1989] 33, 34) is based on the wrong assumption that Alcinous is the same
person as Galen's teacher Albinus and seems unfounded. Cf. [Bobzien 2002a].
65
Pace Barnes [1985]; cf. my [2002a], [2002b].
66
Cf. Gal.Inst.Log.5.1.

17
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

the hypothetical syllogism’, as they called it, for contraries with intermediates, and
expressed their first premises grammatically as disjunctions. 67

There are no signs that Alexander was acquainted with Galen’s suggestions or with the
alternatives chosen by the later commentators. Still, he was aware that there was a
problem with the third type of indemonstrables and that there was a debate about it. We
have no evidence that Alexander himself had a firm view on the matter. However, he is
a valuable witness for the transitional period in which some Peripatetics, and
presumably some Platonists, attempted both to retain all five Stoic types of
indemonstrables, and to pass them all off as Peripatetic (or Platonist), as far as their
origin, character, and justification are concerned. The four Early Peripatetic
hypothetical syllogisms are thus augmented by a fifth. I consider the relevant passages
in Alexander’s commentaries in turn.

3.1 The Prior Analytics commentary:

In the above-discussed passage (3)(ii) Alexander lists three types of hypothetical
arguments that Aristotle could have meant by “the other arguments that conclude from a
hypothesis”:

(3)(ii) He would mean the hypothetical <arguments> by means of that-which-

connects, which is also called a conditional, and the co-assumption, and the ones
by means of that-which-divides, that is disjunction, and perhaps the ones by
means of the negation of a conjunction. (Alex.An.Pr.390.3-6)

The first of these would encompass the first and second Stoic indemonstrables, the
second the fourth and fifth Stoic indemonstrables, and the third the third Stoic
indemonstrables. Two things are noteworthy in (3)(ii): First, for the conditional and
disjunction, Alexander first uses his Peripatetic term and then glosses it by the Stoic
one; but he presents only one expression (the Stoic 'negation of conjunction') for the
third case. This suggests that either there was no terminological Peripatetic counterpart
available to Alexander, or, if there was, it was referred to with the same or a similar
expression as the Stoic. 68 Second, this case is added with the tentative ‘and perhaps’ (ἢ
καὶ). Thus Alexander seems to have some reservations as to whether Aristotle would
have had arguments with negative conjunctions in mind.

The long passage An.Pr.262.28-265.5, (discussed in part in Section 1.1), in which

Alexander attempts to show that the Stoics indemonstrables are Aristotelian syllogisms
from a hypothesis, both confirms these points, and makes them clearer. At the
beginning of the passage Alexander lists in Stoic terminology the three mode-forming
premises used for the five indemonstrables: “the mode-forming <premise/assertible>
being either a conditional or a disjunction or a conjunction” 69 (An.Pr.262.31). Then he

67
E.g. Philop.An.Pr.245.3-23; [Ammon].An.Pr.68.23-41; see also Bobzien [forthcoming].
68
I assume that the early Peripatetics didn’t have an expression for a counterpart to the third
indemonstrables, since they didn’t have such a counterpart; and that the Peripatetics in the 1st century
BCE (e.g. mentioned by Galen Inst.Log.7.2) may have taken over Stoic terminology with Stoic theory.
69
τοῦ τροπικοῦ ἢ συνημμένου ὄντος ἢ διεζευγμένου ἢ συμπεπλεγμένου (Alex.An.Pr.262.31). One might
expect ‘negative conjunction’ rather than ‘conjunction’ here. However, since for Alexander the negative
conjunction is a type of conjunction (whereas the Stoics have ‘negation of a conjunction’, classified as
negation), from Alexander’s perspective, what he says is correct, if unspecific.

18
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

shows that the syllogisms with conditional premise and those with disjunctive premise
fit Aristotle’s description of syllogisms from a hypothesis: the former correspond to the
first and second indemonstrables, the latter to the fourth and fifth ones. (Here again, he
glosses his own expressions τὸ συνεχές and τò διαιρετικóν by the Stoic τò συνημμένον
and τò διεζευγμένον.) Third, he turns to the syllogisms from a negative conjunction. I
quote the passage in full:

(11) But (15) also in the case of the <mode> from a negative conjunction <can it
be shown that it belongs to Aristotle’s syllogisms from a hypothesis>, if, that is,
this mode differs from the previously discussed ones, and is not the same as that
by means of a conditional which begins with an affirmation and ends in a
negation, such as ‘if A, then not B’. 70 For in the case of these <syllogisms>, too,
if the co-assumption requires proof, it is to be proved by means of a categorical
syllogism. For example: ‘it is not both the case that living pleasantly is the goal
(20) and virtue is choiceworthy by itself. But virtue is choiceworthy by itself.
Hence it is not the case that living pleasantly is the goal.’ For that which has
been co-assumed, i.e. ‘virtue is choiceworthy by itself’, is proved through a
categorical syllogism. For example, … (25) … ‘Hence virtue is choiceworthy in
itself.’ But if the same thing were assumed hypothetically in the following form:
‘if pleasure is the goal, then it is not the case that virtue is choiceworthy in
itself’, then the consequence would be proved through a syllogism of this kind:
… . (Alex.An.Pr.264.14-31)

Alexander here considers two possibilities for the syllogism from a negative
conjunction: either (i) it is different from the conditional and disjunctive hypothetical
syllogisms or (ii) it is the same as one of the conditional modes. Alexander does not
commit himself to either possibility. Instead, he demonstrates that either way the
syllogism could be seen as a case of an Aristotelian syllogism from a hypothesis –
showing which is after all the purpose of An.Pr.262-4. It is likely that both options (i)
and (ii) were discussed at Alexander’s time. (i) implies the introduction of a Peripatetic
correlate to the third indemonstrables without the claim of Peripatetic ancestry; (ii)
enables the Peripatetics to argue that the hypothetical syllogism with a negative
conjunction is one of those syllogisms the Peripatetics already accepted (namely a type
of conditional hypothetical syllogism); it is just worded differently (and wording
matters to the Stoics, but not the Peripatetics). 71 Since Alexander does not commit
himself to either view, and is rather tentative in his statements about them, 72 it seems
that at the time there was no Peripatetic standard solution to the problem available. This
notwithstanding, passage (11) sports some interesting features:

We consider option (ii) first: the indemonstrable mode from a negative conjunction (or
negation of a conjunction, as the Stoics would say) is the same as a mode by means of a
conditional. Alexander’s description of the mode as “that by means of a conditional
which begins with an affirmation and ends in a negation” suggests that he assumes that
there are several different modes of the hypothetical syllogisms with a conditional
70
Or: ‘if the first, then not the second’ – the description of the conditional is entirely in Stoic terms,
hence so might have been the illustration.
71
A third possibility, one actually taken by later commentators, would be to add it and equate it with
something else from Aristotle’s logic, such as the case of contraries with intermediates.
72
Cf. the ‘ifs’ at Alex.An.Pr.264.15 and 264.26.

19
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

hypothetical premise. Alexander here uses the Greek word ‘mode’ (τρóπος) not in the
Stoic, but in an Aristotelian or Peripatetic sense: one syllogistic figure has several
modes. The conditional and disjunctive hypothetical syllogisms would be (the
equivalent to) figures, in parallel to Aristotle’s categorical syllogistic. Alexander’s
specification ‘such as “if A, then not B”’ suggests that option (ii) might assume there to
be eight different modes in the ‘conditional figure’: four of type modus ponens, four of
type modus tollens, obtained by drawing out all possible combinations with affirmative
and negative antecedent and consequent propositions. At Alexander’s time, there was a
similar Peripatetic use of ‘mode’ and ‘figure’ for wholly hypothetical syllogisms. 73
Moreover, we know of such a classification of modus ponens and modus tollens
arguments from Boethius and from the Anonymous Scholium Waitz. 74 These two later
texts don’t discuss hypothetical syllogisms with a negative conjunction as hypothetical
premise. So conceivably there was a faction of Peripatetic logicians that integrated the
third Stoic indemonstrable into Peripatetic hypothetical syllogistic by equating it to the
mode of the form ‘if A, not B; A; therefore not B’. 75 Note that the classification of ‘if A,
then not B’ as mode of a conditional hypothetical syllogism differs from that of the
early Peripatetics, who would have considered ‘if A, then not B’ as a – non-standardly
expressed – dividing (or disjunctive) proposition. 76

What about option (i)? Who was the faction of Peripatetics (or other philosophers) that
believed that the hypothetical syllogisms with a negative conjunction differ from the
ones with a conditional or with a disjunction that Alexander discussed beforehand?
They would be either the 1st century BCE Peripatetics who adopted the Stoic
indemonstrables; or alternatively, philosophers like Galen, who classified the negative
conjunctive hypothetical propositions as among those that indicate a kind of conflict
(μάχη), and thus among the other group of hypothetical syllogisms which the early
Peripatetics distinguished. In the latter case, these hypothetical syllogisms, while being
an independent kind, could still be subsumed under the early Peripatetic categorization
of ‘connection’ and ‘division’, which Galen called ἀκολουθία and μάχη, and thus, if
desired, could be reclaimed as Peripatetic in origin.

3.2 The Topics commentary:

One of the Topics passages discussed above in Section 1.2 shows in a different way that
Alexander had not yet found a place for the third indemonstrables:

(10) And the proof that is fitting for the contraries without intermediates is rather
the one by means of the so-called fifth indemonstrable, which is the one that
concludes from that-which-divides and the contradictory to one of the
<components> in that-which-divides the remaining <component>. But for the
contraries with intermediates the <proof> by means of the fourth <indemonstrable
is fitting>, which is the one that refutes from that-which-divides and one of the
<components> in that-which-divides the other <component>.

73
See my [2000].
74
See my [2002c].
75
Later Aristotle commentators generally use the expression ‘mode’ (τρóπος) differently: i.e. to denote
the (by then) five different forms of hypothetical syllogisms: e.g. Philop.An.Pr.244.1-246.14,
Anon.Log.etQuadr.38, Heiberg. (On Log.etQuadr.38 cf. Barnes [2002].)
76
See my [2002a], [2002b].

20
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

We saw in Section 1.2 that Alexander uses the pattern of the fourth and fifth
indemonstrables to describe how, by using them, one can obtain proofs. We noted that
we would have expected him saying that the contradictories without intermediates fit
the pattern of the fifth and the fourth indemonstrables.

Now we have a closer look at Alexander’s suggestion that the contraries with
intermediates are such that the pattern of the fourth indemonstrables is fitting for them.
The connection appears to be roughly this: when two terms are contraries with
intermediates, then they are in conflict (μάχη), since they cannot both hold of the same
thing. The hypothetical premises or propositions that express conflict are the dividing
(or disjunctive) ones. The inference patterns built on dividing (or disjunctive)
hypothetical premises/propositions are those of the fourth and fifth indemonstrables.
The pattern of the fifth indemonstrables does not ‘fit’ the contraries with intermedicates,
since one cannot, for example, infer from the conflict of black and white that if this is
not white, than it is black. It could be grey or yellow. But the pattern of the fourth
indemonstrable does ‘fit’: we can safely infer from the conflict of black and white that if
this is white it is not black. So far, so good.

We are left with two problems: Alexander’s omission of the fact that the pattern of the
fourth indemonstrables also fits the contraries with intermediates (and contradictories,
of course). And the difficulty that the hypothetical propositions in the fourth and fifth
indemonstrables are expressed in disjunctions, whereas disjunctions are not a suitable
way for expressing the conflict of contraries with intermediates. We just don’t say “this
is either black or white” when we wish to express that particular kind of conflict.

The third indemonstrable is not at all considered by Alexander in the passages quoted
from his Topics commentary (nor anywhere in the commentary), although it would lend
itself perfectly to arguments using contraries with intermediates. Some later
Peripatetics or Platonists solved both of Alexander’s problems, by assigning both the
pattern of the fourth and the fifth indemonstrables to the contraries without
intermediates, and by assigning the pattern of the third indemonstrable to contraries
with intermediates. 77 The negative conjunction with component sentences that share the
same subject term is the perfect grammatical vehicle for expressing contraries with
intermediates. ('This is not both black and white. But it is black. Hence it is not white.')
Alexander equipped these later philosophers with the connection of Aristotle’s
classification of opposites (from the Categories) with (i) the passages from Aristotle’s
Topics and with (ii) the Stoic indemonstrables. He thus laid the foundation for this
later, more coherent, theory.

4. Conclusion
Alexander is the first known ancient author who both suggested that Aristotle was
aware of the inference patterns the Stoics encapsulated in their theory of
indemonstrables, and attempted to provide evidence for this fact by drawing a
connection between the indemonstrables and selected passages from Aristotle’s
Organon. Furthermore – presumably in order to support his view that Aristotle had
considered such inference patterns, and to establish a Peripatetic theory of them (a
hypothetical syllogistic) – Alexander introduced a logical vocabulary that is based on
77
See my [forthcoming].

21
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

Aristotelian and early Peripatetic terminology. He modified Aristotelian and early

Peripatetic terms and their use in such a way that they fit a theory of syllogisms derived
from the Stoic indemonstrables. The resulting elements of a hypothetical syllogistic are
Peripatetic both in spirit and in nomenclature. Alexander did not manage to integrate the
third Stoic indemonstrables into the Peripatetic system. But he provided the foundations
for what later became the standard way of doing so.

The importance of Alexander in the development of a Peripatetic (and Platonist)

propositional logic is thus twofold: first he does no longer condone the eclectic method
used by earlier Peripatetics, who unabashedly (and sometimes without
acknowledgement) took over parts of Stoic logic to complement Aristotelian logic.
Instead, now, elements of Stoic logic that are integrated in Peripatetic theory have to be
shown to have their origin in some Aristotelian (or early Peripatetic) thought. Second,
Alexander paved the way for a full Peripatetic/Platonist hypothetical syllogistic by
drawing a number of important, if sometimes somewhat far-fetched, connections
between Stoic logic and Aristotle’s Organon; connections on which later Aristotle
commentators were able to build.

Acknowledgements
I am indebted to Fabio Acerbi and István Bodnár for helpful comments on a draft of this
paper.

References
Barnes, J. (1984). ‘Terms and Sentences: Theophrastus on Hypothetical syllogisms”,
Proceedings of the British Academy 69: 279-326.
Barnes, J. (1985). ‘Theophrastus and Hypothetical Syllogistic’, in J.Wiesner (ed.), Aristoteles
Werk und Wirkung, vol.1 (Berlin/New York: De Gruyter) 557-576.
Barnes, J. (2002). ‘Syllogistic in the Anon Heiberg’, in K.Ierodiakonou (ed.), Byzantine
Philosophy and Its Ancient Sources (Oxford: OUP).
Bobzien, S. (1996). ‘Stoic Syllogistic’, Oxford Studies in Ancient Philosophy 14: 133-92.
Bobzien, S. (1997). ‘Stoic Hypotheses and Hypothetical Argument’, Phronesis 42: 299-312.
Bobzien, S. (1999). ‘Stoic Logic’, in K. Algra et al. (eds), The Cambridge History of
Hellenistic Philosophy (Cambridge: CUP) 92-157.
Bobzien, S. (2000). ‘Wholly hypothetical Syllogisms’, Phronesis 45: 87-137.
Bobzien, S. (2002a). ‘The Development of Modus Ponens in Antiquity’, Phronesis 47: 359-394.
Bobzien, S. (2002b). ‘Pre-Stoic hypothetical syllogistic in Galen’ in V.Nutton (ed.) The
Unknown Galen, Bulletin of the Institute of Classical Studies, S77: 57-72.
Bobzien, S. (2002c). ‘A Greek Parallel to Boethius’ De Hypotheticis Syllogismis’, Mnemosyne
55: 285-300.
Bobzien, S. (2004). 'Hypothetical Syllogistic in Galen - Propositional logic off the rails?'
Rhizai: Journal for Ancient Philosophy and Science 2: 57-102.
Bobzien, S. (forthcoming). ‘The making of a Platonist Hypothetical Syllogistic’.
Frede, M. (1974). Die stoische Logik (Göttingen: Vandenhoeck&Ruprecht).

22
 Susanne Bobzien: Alexander of Aphrodisias on Aristotle’s theory of the Stoic indemonstrables

Frede, M. (1975). ‘Stoic vs. Peripatetic Syllogistic’, Archiv für Geschichte der Philosophie 56:
99-124.
Goulet, R. (1978). ‘La classification des propositions simples’, in J.Brunschwig (ed.), Les
Stoïciens et leur Logique (Paris: Vrin) 191-222.
Ierodiakonou, K. (1990). ‘Rediscovering some Stoic arguments’, in P.Nicolacopoulos (ed.),
Greek Studies in the Philosophy and History of Science (Dordrecht: Kluwer) 137-148.
Maroth, M. (1979). 'Die hypothetischen Syllogismen’, Acta Antiqua 27: 407-36.
Maroth, M. (1989). Ibn Sina und die peripatetische "Aussagenlogik" (Leiden: Brill).
Mignucci, M. (1993). ‘The Stoic Themata’, in K.Doering and Th.Ebert (eds) Dialektiker und
Stoiker (Stuttgart: Franz Steiner) 217-238.
Mueller, Ian (1969). ‘Stoic and Peripatetic Logic’, Archiv für Geschichte der Philosophie 51:
173–187.
Mueller, I. (2006). Alexander of Aphrodisiac: on Aristotle’s Prior Analytics 1.23-31 (London:
Duckworth).
Schiefsky, M. (This volume).
Speca, A. (2001). Hypothetical Syllogistic and Stoic Logic (Leiden: Brill).
Street, T. (2001). 'The "Eminent Later Scholar" in Avicenna's Book of the Syllogism', Arabic
Sciences and Philosophy 11: 205-218.
Striker, G. (1979). ‘Aristoteles über Syllogismen "aufgrund einer Hypothese"’, Hermes 107: 33-
50.
Striker, G. (2009). Aristotle, Prior Analytics: Book I (Oxford: OUP).

23